多模态大模型开展开源数据研判的可行性研究
Feasibility Study of Multimodal Macromodeling for Open Source Data Research and Adjudication
刘伟旻 1王斌2
作者信息
- 1. 华南理工大学,广州 510641;广州汇智通信技术有限公司,广州 510631
- 2. 广州汇智通信技术有限公司,广州 510631
- 折叠
摘要
随着文本大模型的发布,逐步形成以文本模态为主流的系统应用,取得一定的成绩,考虑到真实世界存在多种模态,文本大模型也逐步向多模态大模型演进.开源数据以复杂、多样、海量为主要特点,传统手段进行开源数据研判存在难度大、模型难收敛、联动困难等问题,利用多模态大模型技术的非结构化治理能力、智能联动分析能力重塑开源数据研判技术路线,具有很强的应用意义.
Abstract
With the release of the text big model,gradually formed to the text modal as the mainstream of the system application,and achieved certain achievements,considering the existence of a variety of modalities in the real world,the text big model is also gradually evolved to the multi-modal big model.Open source data is mainly characterized by complexity,diversity,and massive volume,and traditional means for open source data research and adjudication have the problems of difficulty,model convergence,and linkage difficulties,etc.It is of strong application significance to reshape open source data research and adjudication technical routes by using the unstructured governance capability and intelligent linkage analysis capability of multimodal big model technology.
关键词
多模态/大模型/人工智能/数据研判Key words
multimodal/big model/artificial intelligence/data adjudication引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据